Ancient DNA | North Africa | Atlantic islands

The Guanches: Berbers who crossed the sea and never sailed again

When Castile arrived in the 15th century, the Canary Islands were home to people with no boats, no metal and a Berber language. Their genomes tell us where they came from, how they lived alone for twelve centuries, and how much of them survives in Canarians today.

3rd c. ADestimated date of the first permanent settlement, from ancient genomes and radiocarbon dates
80%of Guanche ancestry traces to Neolithic Morocco (qpAdm, Serrano et al. 2023)
42% vs 16%indigenous share of maternal (at least) versus paternal lineages in Canarians today, from ancient DNA (Fregel et al. 2009)

Hero image: AI generated illustration of Guanche herders on Tenerife watching Castilian ships approach, with Mount Teide behind them. It is not a reconstruction of real individuals.

In 1402, Norman adventurers sailing for the Crown of Castile landed on Lanzarote and found people who should not have been there. The islands lie about 100 kilometres off the coast of Africa, yet their inhabitants had no seagoing boats, did not travel between islands, and lived much as Neolithic farmers had lived on the mainland thousands of years earlier. Europeans would argue about who they were for the next five centuries.

Atlantis survivors, a lost Viking colony, the last surviving Cro-Magnons of Europe: the Guanches have attracted almost every romantic theory imaginable. Ancient DNA has now settled the question. They were Berbers, North African people who reached the islands around the 3rd century AD and then lived in near total isolation until the Castilian conquest. The story DNA tells is in some ways stranger than the myths. Each island became its own small world, some populations shrank close to collapse, and when the Europeans came, the native people were not wiped out. They were absorbed, mostly through their women.

A people of seven islands

Strictly speaking, Guanche is the name of the native people of Tenerife only. The other islands had their own names: the Majos of Lanzarote and Fuerteventura, the Canarios of Gran Canaria, the Gomeros of La Gomera, the Bimbapes of El Hierro and the Benahoaritas of La Palma. Today Guanche is commonly used for all of them, and we follow that convention here.

Their way of life was remarkably uniform. They herded goats and sheep, grew barley and wheat, ate gofio (toasted grain flour that Canarians still eat today), and lived in natural caves or stone houses. They had no metal, so their tools were made of stone, obsidian, bone and wood. On Tenerife the dead were carefully mummified, wrapped in goatskins and placed in burial caves. Rock inscriptions in a Libyco-Berber script and the few words recorded by the conquerors point clearly to a language related to Berber (Amazigh), the native language family of North Africa.

Political organisation differed from island to island. Tenerife was divided into nine kingdoms ruled by kings called menceys. Gran Canaria had two main chiefdoms. The Castilian conquest took almost a century: Lanzarote fell in 1402, Fuerteventura and El Hierro soon after, while Gran Canaria resisted until 1483, La Palma until 1493 and Tenerife until 1496. On Tenerife the Guanches won the First Battle of Acentejo in 1494, one of the worst defeats of the Castilian conquest, before epidemic disease, famine and a second campaign broke their resistance.

Myths: Atlantis, Vikings and blond Cro-Magnons

Early chroniclers described some islanders as tall and fair, and that detail fed centuries of speculation. In the 19th century, anthropologists measuring Guanche skulls decided they resembled the Upper Palaeolithic Cro-Magnon skulls of France, and the idea of the Guanches as a surviving Stone Age European race became popular. Other writers linked them to Atlantis, to Vikings, to Celts or to lost Mediterranean sailors.

If the Guanches were survivors of Palaeolithic Europe, their genomes should carry European hunter-gatherer ancestry. If they were Berbers, they should look like ancient and modern North Africans. These are testable predictions.

What ancient DNA has found

The first genome-wide study, by Rodriguez-Varela and colleagues in 2017, sequenced Guanches from Tenerife and Gran Canaria dated between the 7th and 11th centuries AD. They were most closely related to modern North Africans of Berber ancestry. The males carried haplogroup E-M183 (a branch of E-M81), the most typical Berber paternal lineage, and part of their ancestry traced back to Neolithic farmers of Anatolian origin.

Fregel and colleagues then sequenced 48 ancient mitochondrial genomes from all seven islands in 2019. Most maternal lineages had a Mediterranean distribution linked to Neolithic migrations into North Africa, and four lineages were found only in the Canaries. Some lineages were more frequent on the islands closest to Africa, which the authors read as evidence of at least two distinct waves of colonisation.

The most complete picture came in 2023, when Serrano, Fregel and colleagues published genome-wide data from 40 individuals from all seven islands, dated between the 3rd and 16th centuries AD. Their main findings were these.

  • A North African origin. The islanders are modelled as about 73 percent Late Neolithic Moroccan ancestry, 7 percent Early Neolithic Moroccan, 13 percent European Bronze Age (Bell Beaker) and 6 percent ancestry related to the East African hunter-gatherer from Mota, a marker of ancient trans-Saharan contact.
  • A settlement around the 3rd century AD. Genetic estimates and radiocarbon dates converge on a colonisation in the first centuries AD.
  • An east-west divide. The western islands (Tenerife, La Gomera, La Palma, El Hierro) carry more Early Neolithic North African ancestry, the eastern islands (Gran Canaria, Fuerteventura, Lanzarote) more European Bronze Age ancestry. The difference was present from the start, pointing to separate founding groups.
  • No gene flow between islands. Each island evolved on its own. The big islands, Tenerife and Gran Canaria, kept fairly stable populations. The smaller and poorer ones did not: on El Hierro the population crashed between the 7th and 12th centuries, and several late individuals from La Gomera, El Hierro and Fuerteventura carry the long runs of homozygosity that come from parents who were close relatives.
  • Typical Berber fathers. Most men carried E-M183, with rarer lineages such as E-M78, T-M184, R-M269 and the sub-Saharan E-M33.

Our analysis 1: who are the Guanches closest to today?

Our datasets contain Global25 coordinates for 34 ancient Canary Islanders: 26 from Serrano et al. 2023, 5 from Rodriguez-Varela et al. 2017 and 3 further individuals (labelled LLD) from the same Global25 compilation. One individual, CAN.009, has only 9.6 percent coverage and sits far from all the others (distance 0.22 to the group average, when all the others sit within 0.06), a typical sign of low coverage noise. We excluded it. We then ranked more than 3,600 modern population averages by their distance to the remaining 33.

Closest modern populations to the Guanches (Global25 distance)0.000.050.100.150.20Jebala, Chefchaouen (Morocco) (n=8): distance 0.01381. Jebala, Chefchaouen (Morocco) (n=8)0.014Morocco E. Middle Atlas Tamazight (n=3): distance 0.01882. Morocco E. Middle Atlas Tamazight (n=3)0.019Algeria Shawiya (n=8): distance 0.02023. Algeria Shawiya (n=8)0.020Morocco Riffian (n=29): distance 0.02074. Morocco Riffian (n=29)0.021Tunisia Chenini (n=18): distance 0.02095. Tunisia Chenini (n=18)0.021Tunisia Tamezret (n=10): distance 0.02236. Tunisia Tamezret (n=10)0.022Algeria Kabyle (n=25): distance 0.02537. Algeria Kabyle (n=25)0.025Moroccan Arab and Berber mixed (n=3): distance 0.02568. Moroccan Arab and Berber mixed (n=3)0.026Tunisia Djerba (n=3): distance 0.02709. Tunisia Djerba (n=3)0.027Morocco Central Atlas Tamazight (n=7): distance 0.027610. Morocco Central Atlas Tamazight (n=7)0.028Tunisia Matmata (n=13): distance 0.027711. Tunisia Matmata (n=13)0.028Tunisia Zraoua (n=15): distance 0.027812. Tunisia Zraoua (n=15)0.028for comparisonMorocco Shilha (n=21): distance 0.0460, rank 2727. Morocco Shilha (n=21)0.046Saharawi (n=27): distance 0.0597, rank 3838. Saharawi (n=27)0.060Andalusian (Spain) (n=33): distance 0.1871, rank 297297. Andalusian (Spain) (n=33)0.187Berber-speakingArabic-speaking North AfricanIberian
Figure 1. The twelve modern populations closest to the Guanche average in Global25, with three reference populations. All twelve are North African. Ten are Berber-speaking groups; the other two are northern Moroccans of largely Berber descent, such as the Arabic-speaking Jebala.

The result is unambiguous. Every one of the twelve closest modern populations is North African: Berber groups from the Moroccan Rif and Middle Atlas, the Shawiya and Kabyles of Algeria, and the Berber villages of southern Tunisia. Mainland Spaniards are not even in the same region of the genetic map, at seven to thirteen times the distance. Interestingly, the closest groups are northern Moroccans, Algerians and Tunisians rather than the southern Moroccan Shilha or the Saharawi, the populations living directly opposite the islands. In the ancestry model described below, the Shilha and the Saharawi carry less European Bronze Age ancestry (7 and 4 percent, against 16 percent in the Guanches and 17 percent in the Riffians), and the Saharawi more ancestry related to sub-Saharan Africa. The Atlantic side of Morocco was probably reshaped by later movements across the Sahara, which the isolated islanders never received.

Our analysis 2: rebuilding the Guanche genome from ancient sources

Next we modelled each Guanche with non-negative least squares (NNLS), using the same four ancient sources as the 2023 study: Late Neolithic Morocco (Kehf el Baroud, about 3000 BC), Early Neolithic Morocco (Ifri n'Amr ou Moussa, about 5000 BC), Bell Beaker Germany and the 4500 year old Mota hunter-gatherer from Ethiopia. The sources are well separated (smallest pole separation 0.22 between Late Neolithic Morocco and Bell Beaker; condition number 6.2), so the model is not exposed to a collinearity trap.

Where Guanche ancestry came from: four-source modelSerrano et al. 2023, qpAdmSerrano et al. 2023, qpAdm: Morocco Late Neolithic 73.3%73%Serrano et al. 2023, qpAdm: Morocco Early Neolithic 6.9%Serrano et al. 2023, qpAdm: Bell Beaker (European Bronze Age) 13.4%13%Serrano et al. 2023, qpAdm: East African hunter-gatherer (Mota) 6.4%This article, G25 (33 genomes)This article, G25 (33 genomes): Morocco Late Neolithic 59.3%59%This article, G25 (33 genomes): Morocco Early Neolithic 19.6%20%This article, G25 (33 genomes): Bell Beaker (European Bronze Age) 15.9%16%This article, G25 (33 genomes): East African hunter-gatherer (Mota) 5.2%AD 330 to 958 (9 genomes)AD 330 to 958 (9 genomes): Morocco Late Neolithic 58.8%59%AD 330 to 958 (9 genomes): Morocco Early Neolithic 18.6%19%AD 330 to 958 (9 genomes): Bell Beaker (European Bronze Age) 17.3%17%AD 330 to 958 (9 genomes): East African hunter-gatherer (Mota) 5.3%AD 1095 to 1385 (23 genomes)AD 1095 to 1385 (23 genomes): Morocco Late Neolithic 59.9%60%AD 1095 to 1385 (23 genomes): Morocco Early Neolithic 19.7%20%AD 1095 to 1385 (23 genomes): Bell Beaker (European Bronze Age) 15.2%15%AD 1095 to 1385 (23 genomes): East African hunter-gatherer (Mota) 5.2%Morocco Late NeolithicMorocco Early NeolithicBell Beaker (European Bronze Age)East African hunter-gatherer (Mota)
Figure 2. Four-source ancestry model for the Guanches. Top bar: the published qpAdm result. Other bars: our Global25 NNLS results for all 33 individuals and for early and late groups. The fit distance of the pooled model is 0.0308.

Our figures agree with the published ones where it matters. North African Neolithic ancestry totals 79 percent in our model against 80 percent with qpAdm, European Bronze Age ancestry 15.9 percent against 13.4, and Mota related ancestry 5.2 percent against 6.4. The split between the two Moroccan Neolithic sources differs: Global25 gives more weight to the Early Neolithic source than qpAdm does. These two sources are related, since the Late Neolithic Moroccans descended partly from the earlier ones, and Global25 has less power than qpAdm to separate them. For the split, the published qpAdm figure should be preferred.

The European Bronze Age component does not mean that Europeans settled the Canaries. This ancestry had already reached North Africa long before, through Bronze Age contacts across the Strait of Gibraltar and later through Punic and Roman settlement. The modern Riffians, for example, carry it too. The colonists simply brought it with them.

The Cro-Magnon test

To test the old Cro-Magnon theory directly, we added a fifth source: Iberian hunter-gatherers, the closest available proxy for the Palaeolithic and Mesolithic people of southwestern Europe. Their share came out at exactly zero. A zero only means something if the model could have detected the ancestry, so we ran a spike-in test, mixing known amounts of hunter-gatherer ancestry into the Guanche average and fitting again with realistic noise. Below about 5 percent, added hunter-gatherer ancestry is mostly absorbed by the Bell Beaker source, which itself carries some; from 7 to 8 percent upward it is detected in almost every run. The honest conclusion is that the Guanches carry no detectable extra European hunter-gatherer ancestry, and certainly nothing like the large share a surviving Palaeolithic population would show. The skulls that looked like Cro-Magnons were Berber skulls.

Our analysis 3: twelve centuries of stability

Individual Guanches, oldest to youngest (G25 NNLS)AD 330 CAN.014CAN.014 (AD 330, coverage 74.93%): Morocco Late Neolithic 60.0%CAN.014 (AD 330, coverage 74.93%): Morocco Early Neolithic 14.8%CAN.014 (AD 330, coverage 74.93%): Bell Beaker (European Bronze Age) 19.8%CAN.014 (AD 330, coverage 74.93%): East African hunter-gatherer (Mota) 5.4%AD 404 CAN.017CAN.017 (AD 404, coverage 98.51%): Morocco Late Neolithic 61.5%CAN.017 (AD 404, coverage 98.51%): Morocco Early Neolithic 14.7%CAN.017 (AD 404, coverage 98.51%): Bell Beaker (European Bronze Age) 19.4%CAN.017 (AD 404, coverage 98.51%): East African hunter-gatherer (Mota) 4.4%AD 515 CAN.018CAN.018 (AD 515, coverage 64.58%): Morocco Late Neolithic 53.5%CAN.018 (AD 515, coverage 64.58%): Morocco Early Neolithic 13.8%CAN.018 (AD 515, coverage 64.58%): Bell Beaker (European Bronze Age) 27.1%CAN.018 (AD 515, coverage 64.58%): East African hunter-gatherer (Mota) 5.6%AD 626 gun012gun012 (AD 626, coverage 37.23%): Morocco Late Neolithic 60.6%gun012 (AD 626, coverage 37.23%): Morocco Early Neolithic 21.9%gun012 (AD 626, coverage 37.23%): Bell Beaker (European Bronze Age) 9.6%gun012 (AD 626, coverage 37.23%): East African hunter-gatherer (Mota) 7.8%AD 713 CAN.034CAN.034 (AD 713, coverage 35.27%): Morocco Late Neolithic 68.6%CAN.034 (AD 713, coverage 35.27%): Morocco Early Neolithic 19.6%CAN.034 (AD 713, coverage 35.27%): Bell Beaker (European Bronze Age) 8.1%CAN.034 (AD 713, coverage 35.27%): East African hunter-gatherer (Mota) 3.7%AD 813 gun011gun011 (AD 813, coverage 96.63%): Morocco Late Neolithic 52.2%gun011 (AD 813, coverage 96.63%): Morocco Early Neolithic 27.0%gun011 (AD 813, coverage 96.63%): Bell Beaker (European Bronze Age) 15.8%gun011 (AD 813, coverage 96.63%): East African hunter-gatherer (Mota) 5.0%AD 884 CAN.027CAN.027 (AD 884, coverage 29.08%): Morocco Late Neolithic 56.4%CAN.027 (AD 884, coverage 29.08%): Morocco Early Neolithic 22.5%CAN.027 (AD 884, coverage 29.08%): Bell Beaker (European Bronze Age) 15.5%CAN.027 (AD 884, coverage 29.08%): East African hunter-gatherer (Mota) 5.5%AD 936 gun008gun008 (AD 936, coverage 24.38%): Morocco Late Neolithic 62.8%gun008 (AD 936, coverage 24.38%): Morocco Early Neolithic 15.5%gun008 (AD 936, coverage 24.38%): Bell Beaker (European Bronze Age) 16.3%gun008 (AD 936, coverage 24.38%): East African hunter-gatherer (Mota) 5.4%AD 958 gun005gun005 (AD 958, coverage 32.95%): Morocco Late Neolithic 53.7%gun005 (AD 958, coverage 32.95%): Morocco Early Neolithic 17.2%gun005 (AD 958, coverage 32.95%): Bell Beaker (European Bronze Age) 24.4%gun005 (AD 958, coverage 32.95%): East African hunter-gatherer (Mota) 4.6%AD 1095 CAN.048CAN.048 (AD 1095, coverage 95.52%): Morocco Late Neolithic 59.0%CAN.048 (AD 1095, coverage 95.52%): Morocco Early Neolithic 24.0%CAN.048 (AD 1095, coverage 95.52%): Bell Beaker (European Bronze Age) 11.4%CAN.048 (AD 1095, coverage 95.52%): East African hunter-gatherer (Mota) 5.7%AD 1099 gun002gun002 (AD 1099, coverage 17.27%): Morocco Late Neolithic 61.7%gun002 (AD 1099, coverage 17.27%): Morocco Early Neolithic 23.1%gun002 (AD 1099, coverage 17.27%): Bell Beaker (European Bronze Age) 9.6%gun002 (AD 1099, coverage 17.27%): East African hunter-gatherer (Mota) 5.6%AD 1133 CAN.036CAN.036 (AD 1133, coverage 93.39%): Morocco Late Neolithic 60.5%CAN.036 (AD 1133, coverage 93.39%): Morocco Early Neolithic 22.7%CAN.036 (AD 1133, coverage 93.39%): Bell Beaker (European Bronze Age) 11.5%CAN.036 (AD 1133, coverage 93.39%): East African hunter-gatherer (Mota) 5.4%AD 1178 LLD-0056LLD-0056 (AD 1178, coverage 12.93%): Morocco Late Neolithic 62.3%LLD-0056 (AD 1178, coverage 12.93%): Morocco Early Neolithic 17.2%LLD-0056 (AD 1178, coverage 12.93%): Bell Beaker (European Bronze Age) 16.2%LLD-0056 (AD 1178, coverage 12.93%): East African hunter-gatherer (Mota) 4.3%AD 1178 LLD-0057LLD-0057 (AD 1178, coverage 24.96%): Morocco Late Neolithic 55.1%LLD-0057 (AD 1178, coverage 24.96%): Morocco Early Neolithic 16.3%LLD-0057 (AD 1178, coverage 24.96%): Bell Beaker (European Bronze Age) 23.7%LLD-0057 (AD 1178, coverage 24.96%): East African hunter-gatherer (Mota) 5.0%AD 1178 LLD-0058LLD-0058 (AD 1178, coverage 19.53%): Morocco Late Neolithic 55.7%LLD-0058 (AD 1178, coverage 19.53%): Morocco Early Neolithic 12.4%LLD-0058 (AD 1178, coverage 19.53%): Bell Beaker (European Bronze Age) 24.8%LLD-0058 (AD 1178, coverage 19.53%): East African hunter-gatherer (Mota) 7.0%AD 1210 CAN.035CAN.035 (AD 1210, coverage 15.6%): Morocco Late Neolithic 60.5%CAN.035 (AD 1210, coverage 15.6%): Morocco Early Neolithic 28.2%CAN.035 (AD 1210, coverage 15.6%): Bell Beaker (European Bronze Age) 10.0%CAN.035 (AD 1210, coverage 15.6%): East African hunter-gatherer (Mota) 1.3%AD 1212 CAN.031CAN.031 (AD 1212, coverage 60.82%): Morocco Late Neolithic 60.4%CAN.031 (AD 1212, coverage 60.82%): Morocco Early Neolithic 21.3%CAN.031 (AD 1212, coverage 60.82%): Bell Beaker (European Bronze Age) 14.0%CAN.031 (AD 1212, coverage 60.82%): East African hunter-gatherer (Mota) 4.3%AD 1213 CAN.038CAN.038 (AD 1213, coverage 95.48%): Morocco Late Neolithic 50.8%CAN.038 (AD 1213, coverage 95.48%): Morocco Early Neolithic 20.3%CAN.038 (AD 1213, coverage 95.48%): Bell Beaker (European Bronze Age) 26.0%CAN.038 (AD 1213, coverage 95.48%): East African hunter-gatherer (Mota) 3.0%AD 1217 CAN.012CAN.012 (AD 1217, coverage 76.49%): Morocco Late Neolithic 66.9%CAN.012 (AD 1217, coverage 76.49%): Morocco Early Neolithic 11.9%CAN.012 (AD 1217, coverage 76.49%): Bell Beaker (European Bronze Age) 15.9%CAN.012 (AD 1217, coverage 76.49%): East African hunter-gatherer (Mota) 5.2%AD 1255 CAN.003CAN.003 (AD 1255, coverage 63.56%): Morocco Late Neolithic 58.9%CAN.003 (AD 1255, coverage 63.56%): Morocco Early Neolithic 23.6%CAN.003 (AD 1255, coverage 63.56%): Bell Beaker (European Bronze Age) 11.6%CAN.003 (AD 1255, coverage 63.56%): East African hunter-gatherer (Mota) 5.9%AD 1262 CAN.041CAN.041 (AD 1262, coverage 24.59%): Morocco Late Neolithic 56.6%CAN.041 (AD 1262, coverage 24.59%): Morocco Early Neolithic 21.0%CAN.041 (AD 1262, coverage 24.59%): Bell Beaker (European Bronze Age) 17.6%CAN.041 (AD 1262, coverage 24.59%): East African hunter-gatherer (Mota) 4.8%AD 1327 CAN.011CAN.011 (AD 1327, coverage 17.65%): Morocco Late Neolithic 52.6%CAN.011 (AD 1327, coverage 17.65%): Morocco Early Neolithic 13.6%CAN.011 (AD 1327, coverage 17.65%): Bell Beaker (European Bronze Age) 25.8%CAN.011 (AD 1327, coverage 17.65%): East African hunter-gatherer (Mota) 7.9%AD 1330 CAN.001CAN.001 (AD 1330, coverage 93.49%): Morocco Late Neolithic 55.6%CAN.001 (AD 1330, coverage 93.49%): Morocco Early Neolithic 25.8%CAN.001 (AD 1330, coverage 93.49%): Bell Beaker (European Bronze Age) 11.7%CAN.001 (AD 1330, coverage 93.49%): East African hunter-gatherer (Mota) 6.9%AD 1330 CAN.005CAN.005 (AD 1330, coverage 97.2%): Morocco Late Neolithic 61.1%CAN.005 (AD 1330, coverage 97.2%): Morocco Early Neolithic 16.9%CAN.005 (AD 1330, coverage 97.2%): Bell Beaker (European Bronze Age) 17.6%CAN.005 (AD 1330, coverage 97.2%): East African hunter-gatherer (Mota) 4.5%AD 1332 CAN.040CAN.040 (AD 1332, coverage 22.43%): Morocco Late Neolithic 71.6%CAN.040 (AD 1332, coverage 22.43%): Morocco Early Neolithic 12.1%CAN.040 (AD 1332, coverage 22.43%): Bell Beaker (European Bronze Age) 8.5%CAN.040 (AD 1332, coverage 22.43%): East African hunter-gatherer (Mota) 7.8%AD 1342 CAN.030CAN.030 (AD 1342, coverage 51.01%): Morocco Late Neolithic 55.9%CAN.030 (AD 1342, coverage 51.01%): Morocco Early Neolithic 29.3%CAN.030 (AD 1342, coverage 51.01%): Bell Beaker (European Bronze Age) 12.8%CAN.030 (AD 1342, coverage 51.01%): East African hunter-gatherer (Mota) 2.0%AD 1344 CAN.004CAN.004 (AD 1344, coverage 65.29%): Morocco Late Neolithic 58.9%CAN.004 (AD 1344, coverage 65.29%): Morocco Early Neolithic 21.0%CAN.004 (AD 1344, coverage 65.29%): Bell Beaker (European Bronze Age) 11.6%CAN.004 (AD 1344, coverage 65.29%): East African hunter-gatherer (Mota) 8.5%AD 1351 CAN.020CAN.020 (AD 1351, coverage 24.24%): Morocco Late Neolithic 54.3%CAN.020 (AD 1351, coverage 24.24%): Morocco Early Neolithic 17.5%CAN.020 (AD 1351, coverage 24.24%): Bell Beaker (European Bronze Age) 24.5%CAN.020 (AD 1351, coverage 24.24%): East African hunter-gatherer (Mota) 3.7%AD 1362 CAN.046CAN.046 (AD 1362, coverage 13.67%): Morocco Late Neolithic 67.0%CAN.046 (AD 1362, coverage 13.67%): Morocco Early Neolithic 19.7%CAN.046 (AD 1362, coverage 13.67%): Bell Beaker (European Bronze Age) 9.0%CAN.046 (AD 1362, coverage 13.67%): East African hunter-gatherer (Mota) 4.2%AD 1368 CAN.008CAN.008 (AD 1368, coverage 95.28%): Morocco Late Neolithic 63.2%CAN.008 (AD 1368, coverage 95.28%): Morocco Early Neolithic 17.7%CAN.008 (AD 1368, coverage 95.28%): Bell Beaker (European Bronze Age) 16.1%CAN.008 (AD 1368, coverage 95.28%): East African hunter-gatherer (Mota) 2.9%AD 1383 CAN.042CAN.042 (AD 1383, coverage 95.11%): Morocco Late Neolithic 65.0%CAN.042 (AD 1383, coverage 95.11%): Morocco Early Neolithic 20.0%CAN.042 (AD 1383, coverage 95.11%): Bell Beaker (European Bronze Age) 8.7%CAN.042 (AD 1383, coverage 95.11%): East African hunter-gatherer (Mota) 6.3%AD 1385 CAN.002CAN.002 (AD 1385, coverage 29.84%): Morocco Late Neolithic 64.2%CAN.002 (AD 1385, coverage 29.84%): Morocco Early Neolithic 17.7%CAN.002 (AD 1385, coverage 29.84%): Bell Beaker (European Bronze Age) 11.1%CAN.002 (AD 1385, coverage 29.84%): East African hunter-gatherer (Mota) 7.0%AD 1534 CAN.045CAN.045 (AD 1534, coverage 48.78%): Morocco Late Neolithic 51.1%CAN.045 (AD 1534, coverage 48.78%): Morocco Early Neolithic 26.0%CAN.045 (AD 1534, coverage 48.78%): Bell Beaker (European Bronze Age) 17.6%CAN.045 (AD 1534, coverage 48.78%): East African hunter-gatherer (Mota) 5.3%Morocco Late NeolithicMorocco Early NeolithicBell Beaker (European Bronze Age)East African hunter-gatherer (Mota)
Figure 3. Individual Guanches sorted by date, from AD 330 to AD 1534. The proportions vary from person to person, partly because of the east-west differences between islands and partly because of statistical noise, but there is no trend over time. CAN.045 (red label) lived after the conquest.

Across more than a thousand years, the ancestry of the islanders barely changes. The nine individuals dated before AD 1000 and the 23 dated between AD 1095 and 1385 give almost identical averages (17.3 and 15.2 percent European Bronze Age ancestry respectively). This is exactly what a population cut off from the mainland should look like: nothing new came in.

The youngest individual, CAN.045, is dated to the early 16th century, a generation or more after the conquest. Genetically, this person is entirely Guanche. When we model CAN.045 with Guanche, Andalusian and West African sources, the Andalusian share is zero. Our spike-in tests on single individuals detect added Spanish ancestry from about 10 percent upward, so CAN.045 had no Spanish parent and very probably no Spanish grandparent. Mixing between natives and settlers was a process that took several generations.

How did they get there without boats?

The question that puzzled Europeans remains partly open. The Guanches had no seagoing boats when the Castilians arrived, yet their ancestors must have crossed at least 100 kilometres of open Atlantic, and then reached all seven islands. Either they lost their seafaring skills after arriving, or someone else brought them.

The date matters here. A settlement around the 3rd century AD falls in the period when North Africa was under Roman rule, and when Mediterranean sailors certainly knew the islands. The Roman author Pliny the Elder describes an expedition to the Canaries sent by King Juba II of Mauretania, a Roman client king, around the turn of the era, and archaeologists have found a Roman workshop on the islet of Lobos, off Fuerteventura, used to produce purple dye from sea snails. One hypothesis is that Berber groups were transported to the islands, voluntarily or not, by Mediterranean powers. DNA cannot prove who owned the ships, but the timing, the mix of ancestries and the separate founding groups of east and west are all compatible with this scenario.

Conquest and absorption

The conquest was brutal. Thousands of islanders were killed, died of disease or were sold as slaves in Spain and Portugal, and the islands were resettled by Castilians, Andalusians, Portuguese and others, along with enslaved Africans brought to work on the sugar plantations. It is often said that the Guanches became extinct. Genetics shows that is wrong.

The clearest evidence comes from uniparental markers. Using ancient remains as the indigenous reference, Fregel and colleagues estimated in 2009 that indigenous lineages make up at least 42 percent (and up to 73 percent, depending on the method) of the maternal (mitochondrial) gene pool of modern Canarians, but only about 16 percent of the paternal (Y chromosome) gene pool. In their words, female lineages survived the conquest with only a moderate decline, while male indigenous lineages were steadily replaced by European ones. Native men were killed, enslaved or excluded; native women married or were taken by settlers, and their lineages survived. It is the same sex-biased pattern seen across the Spanish Americas, and the Canaries were its first laboratory.

At the genome-wide level, Serrano and colleagues model modern Canarians as 79.7 percent Spanish, 17.8 percent indigenous Canarian and 2.6 percent sub-Saharan African. Rodriguez-Varela and colleagues estimated that the inhabitants of Gran Canaria inherit roughly 16 to 31 percent of their ancestry from the Guanches.

Our analysis 4: island by island

We modelled the modern population averages of each island (Moriopoulos 2026 collection) with three sources: the pre-conquest Guanche average, Andalusians and Yoruba as a West African proxy. To check how much the choice of the Spanish source matters, we repeated the model with Castilian, Extremaduran and Galician proxies.

Indigenous (Guanche-like) ancestry in Canary Islanders today, east to west0%5%10%15%20%25%30%qpAdm, Serrano et al. 2023: 17.8%Lanzarote: Guanche-like 20.8% (range 19.6 to 21.9% with four Iberian proxies), Iberian 77.5%, sub-Saharan 1.8%Lanzarote21% (+1.8% sub-Saharan)Fuerteventura: Guanche-like 21.9% (range 20.9 to 23.0% with four Iberian proxies), Iberian 77.0%, sub-Saharan 1.1%Fuerteventura22% (+1.1% sub-Saharan)Gran Canaria: Guanche-like 11.0% (range 9.8 to 12.2% with four Iberian proxies), Iberian 86.6%, sub-Saharan 2.4%Gran Canaria11% (+2.4% sub-Saharan)Tenerife: Guanche-like 10.9% (range 9.6 to 12.0% with four Iberian proxies), Iberian 88.1%, sub-Saharan 1.0%Tenerife11% (+1.0% sub-Saharan)La Gomera: Guanche-like 20.2% (range 19.0 to 21.2% with four Iberian proxies), Iberian 77.0%, sub-Saharan 2.8%La Gomera20% (+2.8% sub-Saharan)La Palma: Guanche-like 10.3% (range 9.0 to 11.5% with four Iberian proxies), Iberian 89.1%, sub-Saharan 0.6%La Palma10% (+0.6% sub-Saharan)El Hierro: Guanche-like 22.5% (range 21.6 to 23.6% with four Iberian proxies), Iberian 77.1%, sub-Saharan 0.4%El Hierro22% (+0.4% sub-Saharan)All (G25 average): Guanche-like 15.0% (range 13.7 to 16.2% with four Iberian proxies), Iberian 84.3%, sub-Saharan 0.8%All (G25 average)15% (+0.8% sub-Saharan)Guanche-like ancestry (model: Guanche + Andalusian + Yoruba)Sub-Saharan AfricanWhiskers: range when the Iberian source is Andalusian, Castilian, Extremaduran or Galician.
Figure 4. Guanche-like ancestry in modern Canarians by island, ordered from east to west. The dashed line marks the archipelago average obtained by Serrano et al. 2023 with qpAdm. Note that in Global25, Guanche ancestry cannot be fully separated from later North African ancestry (see text).
IslandGuanche-likeRange (4 Iberian proxies)IberianSub-SaharanFit distance
Lanzarote20.8%19.6 to 21.9%77.5%1.8%0.0082
Fuerteventura21.9%20.9 to 23.0%77.0%1.1%0.0084
Gran Canaria11.0%9.8 to 12.2%86.6%2.4%0.0096
Tenerife10.9%9.6 to 12.0%88.1%1.0%0.0104
La Gomera20.2%19.0 to 21.2%77.0%2.8%0.0087
La Palma10.3%9.0 to 11.5%89.1%0.6%0.0124
El Hierro22.5%21.6 to 23.6%77.1%0.4%0.0084

Over the whole archipelago our estimate is 15.0 percent (13.7 to 16.2 percent depending on the Spanish source), close to the published 17.8 percent. The island pattern is striking, and it follows a historical fault line. Gran Canaria, Tenerife and La Palma, conquered directly by the Crown (the islas de realengo) and quickly turned into sugar and wine economies that attracted large numbers of settlers, sit around 10 to 11 percent. Lanzarote, Fuerteventura, La Gomera and El Hierro, the islas de senorio held by feudal lords, with poorer land and less immigration, reach 20 to 23 percent.

Our Gran Canaria figure (11 percent) is lower than the 16 to 31 percent range of Rodriguez-Varela and colleagues. Their estimate was based on only a handful of ancient genomes and on a different modern sample; ours depends on the Spanish proxy and on how well the modern average represents the island. Both approaches agree that the indigenous contribution is substantial.

A collinearity trap: Guanche or Moroccan?

There is an important limit to these numbers. The Guanche average and the modern Riffian Berbers are only 0.020 apart in Global25, closer than many pairs of neighbouring European populations. When we offer both as sources in the same model, the condition number jumps from 7 to 61 and the results become unstable: on Lanzarote, for example, most of the North African ancestry moves from the Guanche source to the Riffian one, while on Fuerteventura it stays Guanche.

This matters because the Canaries also received North Africans after the conquest. Spanish raids on the African coast in the 15th and 16th centuries brought many Berber captives, the so-called moriscos, to Lanzarote and Fuerteventura in particular. Global25 cannot tell their descendants from those of the Guanches. The figures in Figure 4 therefore measure North African ancestry of Guanche or similar origin, and on the eastern islands part of it may come from the mainland after 1402. Separating the two would require haplotype-based methods applied to full genomes.

The African thread and the controls

Sub-Saharan ancestry ranges from 0.4 percent on El Hierro to 2.4 percent on Gran Canaria and 2.8 percent on La Gomera, two islands where sugar plantations used enslaved African labour. These small values are reliable: our spike-in tests recover West African ancestry from about 1 percent. As a control, we ran the same model on mainland Spanish and Portuguese populations. Castilians, Galicians, Valencians, Murcians, Basques, Asturians and Cantabrians receive exactly zero Guanche ancestry; Extremadurans and Portuguese receive about 2 percent, below the detection floor of about 3 percent we measured for this source by spike-in, and probably reflecting old North African ancestry on the Iberian mainland itself.

Conclusion: absorbed, not extinct

The Guanches were not Atlanteans, Vikings or the last Cro-Magnons. They were Berbers, descendants of the Neolithic farmers and herders of Morocco, who reached the Canaries around the 3rd century AD, perhaps on ships that were not their own. For twelve centuries each island lived in isolation, with its own founders and its own fate, from the stable populations of Tenerife and Gran Canaria to the near collapse of El Hierro.

The conquest destroyed their societies but not their descendants. Around one sixth of the genome of the average Canarian, and at least two fifths of maternal lineages, still go back to the people the Castilians found on the islands in the 15th century. The Guanches are not a vanished people. They are the grandmothers of the Canary Islands.

Global25 coordinates used in this article

The Guanche average combines 32 pre-conquest individuals from Serrano et al. 2023 and Rodriguez-Varela et al. 2017 (CAN.009 excluded for low coverage). Modern Canarian averages are from the Moriopoulos 2026 collection; reference populations from Global25 by Davidski. Paste them into the Vahaduo tool to run your own models.

Guanche_Canary_Islands_avg_AD330-1385_(n=32),-0.032937,0.133606,-0.001343,-0.067638,0.034131,-0.033066,-0.028892,0.006231,0.071162,0.033418,0.007155,-0.006496,0.017867,-0.017796,0.021079,-0.014419,0.002168,-0.02052,-0.042537,0.008782,-0.014147,-0.036717,0.021688,-0.005885,0.004277
Guanche_CAN.045_AD1534_post-conquest,-0.042115,0.123895,-0.000754,-0.060078,0.040315,-0.049085,-0.035722,-0.001846,0.074856,0.028611,0.002111,-0.014387,0.023637,-0.017478,0.026465,-0.008221,0.011604,-0.015456,-0.052542,0.013506,-0.006863,-0.041671,0.02884,0.007712,0.000838
Morocco_LN.SG,0.019919,0.145729,-0.00264,-0.093186,0.040162,-0.047412,-0.021739,0.004615,0.081094,0.035901,0.006739,0.013188,0.014123,-0.018717,0.000746,-0.001723,0.015842,-0.006462,-0.033121,0.019072,-0.010169,-0.023123,0.008196,-0.010664,0.000179
Morocco_EN.SG,-0.17358,0.092413,-0.026398,-0.082365,0.030621,-0.060241,-0.079434,0.020538,0.151756,0.005376,0.020866,-0.025702,0.074776,-0.045966,0.067182,-0.032551,0.013364,-0.05701,-0.149455,0.032203,-0.040054,-0.117223,0.081529,-0.009519,0.021256
Germany_BellBeaker,0.126736,0.131423,0.056425,0.045008,0.034213,0.015589,0.001386,0.002825,0.005261,0.004726,-0.002139,0.006605,-0.00995,-0.011973,0.014794,0.008467,0.00112,0.001616,0.002289,0.005934,0.005279,0.003982,-0.002678,-0.000976,-0.000991
Ethiopia_4500BP,-0.517895,0.040621,-0.003017,-0.000646,-0.002154,-0.009761,0.057813,-0.049383,0.089172,-0.084375,-0.013803,-0.014987,-0.036571,-0.001651,0.02348,-0.030628,0.022817,0.041047,0.005279,-0.008629,0.00262,0.001855,-0.006039,0.001205,0.002515
Spanish_Andalucia,0.108132,0.146076,0.035707,-0.003434,0.045256,-0.005857,-0.00193,0.00266,0.024715,0.031162,-0.002154,0.007272,-0.013207,-0.011408,0.010243,-0.002498,-0.006176,-0.0011,-0.004261,-0.002133,0.002929,-0.001302,-0.001972,-0.005334,0.001885
Yoruba,-0.630062,0.062501,0.022113,0.016708,0.000503,0.012474,-0.044417,0.047767,-0.048881,0.032769,0.004621,0.00079,0.023056,0.000951,0.012523,-0.009607,0.007076,0.000449,0.006022,-0.00299,0.001554,0.002316,-0.001759,-0.000471,-0.000425
Berber_Morocco_Riffian_(n=29),-0.021744,0.139198,-0.002549,-0.068053,0.030011,-0.031842,-0.022634,0.004591,0.061914,0.029906,0.005023,-0.006222,0.017193,-0.014351,0.015154,-0.012838,-0.001407,-0.018012,-0.038624,0.008327,-0.013188,-0.035216,0.022061,-0.005431,0.004819
Spanish_Canarias_(n=19),0.081294,0.145702,0.030269,-0.012903,0.037837,-0.006224,-0.006494,0.000814,0.031604,0.029052,-0.000496,0.005261,-0.007175,-0.013617,0.011022,-0.002198,-0.002738,-0.003487,-0.010843,0.000987,0.000604,-0.00643,0.001641,-0.003418,0.001702
Spanish_Canarias_El_Hierro_(n=34),0.073818,0.142353,0.028706,-0.019579,0.040107,-0.007932,-0.008204,0.005056,0.033097,0.029528,0.000993,0.004156,-0.006226,-0.013074,0.014566,-0.004855,-0.003651,-0.006916,-0.014847,0.000993,-0.001846,-0.009769,0.002987,-0.001258,0.000768
Spanish_Canarias_Fuerteventura_(n=32),0.069041,0.141825,0.029498,-0.01615,0.041027,-0.010345,-0.007513,0.00318,0.034833,0.029687,0.001822,0.002758,-0.006174,-0.013225,0.012083,-0.001371,-0.001569,-0.00325,-0.012833,0.003638,-0.001883,-0.010422,0.006197,-0.004839,0.002848
Spanish_Canarias_Gran_Canaria_(n=117),0.075435,0.142018,0.031955,-0.011788,0.040286,-0.004643,-0.006219,0.004761,0.026805,0.029924,0.001313,0.00364,-0.007959,-0.013294,0.01195,0.000016,-0.001567,-0.002322,-0.007483,0.001222,0.000673,-0.006029,0.001462,-0.002115,-0.00005
Spanish_Canarias_La_Gomera_(n=76),0.059547,0.141266,0.029564,-0.017068,0.039242,-0.007196,-0.01069,0.003449,0.030875,0.029755,0.000891,0.002587,-0.005639,-0.012417,0.011333,-0.000926,-0.000722,-0.005264,-0.012247,0.000538,-0.001284,-0.009806,0.003376,-0.004014,0.001654
Spanish_Canarias_La_Palma_(n=34),0.089083,0.144325,0.036115,-0.008198,0.042795,-0.005201,-0.003829,0.001873,0.031148,0.031725,0.001065,0.002627,-0.008797,-0.013645,0.012307,0.000335,0.000441,-0.003708,-0.009072,0.000478,0.001971,-0.007426,0.00203,-0.00039,-0.000261
Spanish_Canarias_Lanzarote_(n=56),0.065895,0.143462,0.028769,-0.016254,0.039298,-0.007834,-0.00904,0.003021,0.033615,0.029984,0.000754,0.00379,-0.004585,-0.013907,0.009786,-0.002031,-0.003139,-0.003527,-0.013176,0.0018,-0.001566,-0.009064,0.003651,-0.002487,0.00232
Spanish_Canarias_Tenerife_(n=64),0.085705,0.144697,0.034177,-0.011214,0.040214,-0.005735,-0.005717,0.002751,0.027614,0.029776,-0.001063,0.004719,-0.008102,-0.012608,0.011602,0.000126,-0.001292,-0.002896,-0.0094,0.00119,0.000084,-0.005814,0.002975,-0.00096,-0.000726

Methods in brief

Global25 scaled coordinates were merged with first occurrence priority from Davidski's population averages, Davidski's modern averages and the Moriopoulos 2026 collection (all averages, no sims); individual ancient genomes came from the dated Global25 ancients file. Scale compatibility was confirmed by comparing the published CanaryIslands_Guanche.SG average with the mean of its five individuals (distance 0.009). Models were fitted by NNLS (scipy) with the sum-to-one constraint enforced as a heavily weighted extra equation. Detection floors were estimated by spike-in tests: known proportions of a source were mixed into a target, random noise of the size of the observed fit residuals was added, and the model was refitted 300 to 500 times. Distances are Euclidean distances on the 25 scaled dimensions.

References

  1. Ancient DNA Serrano JG, Ordonez AC, Santana J, et al., Fregel R (2023). The genomic history of the indigenous people of the Canary Islands. Nature Communications 14. doi.org/10.1038/s41467-023-40198-w
  2. Ancient DNA Rodriguez-Varela R, Gunther T, Krzewinska M, et al. (2017). Genomic analyses of pre-European conquest human remains from the Canary Islands reveal close affinity to modern North Africans. Current Biology 27(21). cell.com/current-biology
  3. Ancient mtDNA Fregel R, Ordonez AC, Santana-Cabrera J, et al. (2019). Mitogenomes illuminate the origin and migration patterns of the indigenous people of the Canary Islands. PLoS ONE 14(3): e0209125. doi.org/10.1371/journal.pone.0209125
  4. Y chromosome Fregel R, Gomes V, Gusmao L, et al. (2009). Demographic history of Canary Islands male gene-pool: replacement of native lineages by European. BMC Evolutionary Biology 9: 181. doi.org/10.1186/1471-2148-9-181
  5. Ancient DNA Fregel R, Mendez FL, Bokbot Y, et al. (2018). Ancient genomes from North Africa evidence prehistoric migrations to the Maghreb from both the Levant and Europe. Proceedings of the National Academy of Sciences 115(26). pmc.ncbi.nlm.nih.gov/articles/PMC6042094
  6. Press University of La Laguna (2023). Insularity and resources shaped the genetic composition of the Canarian aborigines. ull.es
  7. Data Global25 PCA coordinates (Davidski) and the Moriopoulos 2026 collection of Global25 averages. eurogenes.blogspot.com